1. Field of the Invention
This invention relates generally to digital broadcasting. More particularly, it relates to a method and apparatus for reliably acquiring orthogonal frequency division multiplexing (OFDM) frame synchronization in a hybrid FM In-Band On-Channel (IBOC) system.
2. Background of Related Art
Various methods of broadcasting signals are known. In particular, a number of attempts have been made to simultaneously broadcast similar copies of the same source material on different frequencies or channels, e.g., both in digital form, and/or one in digital form and one in analog form. This technique can be advantageously employed for non-commercial broadcasts such as cellular phone transmissions, e.g., to provide for backward compatibility and/or signal redundancy, as well as for commercial broadcasts such as digital television (DTV) or radio services such as FM.
For instance, FIG. 5 depicts a conventional method of simultaneously broadcasting two digital copies of the same source material along with an older standard analog waveform. In the shown example, the broadcast includes two digital broadcast signals A and B and one analog broadcast signal C transmitted in the same band on the same channel, otherwise known in the United States as an In-Band On-Channel Digital Audio Broadcast (IBOC DAB) or hybrid IBOC.
FIG. 5 shows a typical frequency spectrum of a IBOC DAB signal including two digital broadcast signals A and B as well as an analog broadcast signal C. The present invention improves the reliability in the detection of the starting point of each frame of data in the digital broadcast signals A, B.
In the disclosed example, the hybrid IBOC DAB signal 600 includes a center band containing a first, analog copy of the source material, an upper band A containing a first digital copy of the same source material, and a lower band B containing a second digital copy of the same source material.
The use of both an analog broadcast signal C and digital broadcast signals A, B allows the capability for both analog and digital receivers to receive and interpret at least one copy of the source material. In many cases, this provides backwards compatibility to analog systems as systems are replaced with more advanced digital counterpart systems.
FIG. 6 shows that both digital broadcast signals A and B (as well as analog broadcast signal C) are conventionally transmitted at the same time. The use of two digital broadcast channels, i.e., the upper band A and the lower band B, provides system robustness and reliability in light of environmental conditions. For instance, both digital broadcast signals A and B can be received by a digital receiver and each received signal checked for errors in transmission using, e.g., a cyclic redundancy check (CRC) or check sum. If one broadcast signal A or B contains errors at any one time, the system can switch over to utilizing the second, error-free channel. This increases the reliability of the system because of the theoretically increased chances of at least one digital signal getting through to the receiver without interference or signal loss due to signal fade. Thus, in theory, the transmission of two copies of the same source material, e.g., one in the upper band A and one in the lower band B, would avoid the destructive effects of some environmental conditions such as frequency selective fading caused by a moving receiver or transmitter. It is typically advantageous to combine digital signals A and B by a soft combination method. A soft combiner combines the two channel codes on the two sidebands. These can be identical codes or preferably complementary punctured codes.
However, in reality, the present inventors have appreciated that one of the channels A or B may be disadvantageously affected by the presence of adjacent stations, causing severe interference with a digital channel A or B, resulting in erroneous detection of the starting point of each frame of digital data.
In particular, frame synchronization is performed so that properly lined up samples can be delivered to the FFT block at the receiver. If this is not done properly, a reliable link cannot be established.
Each frame of data in the OFDM signal may be, e.g., 512, 1024, 2048, etc. bits in length. A conventional technique for ‘marking’ the beginning point of the OFDM data frame is called “prefix extension”.
Prefix extension relates to the cyclical extension of a number of bits, copied either from the end of the data frame and/or from the beginning of the data frame, and adding the same to the opposite end of the data frame. For instance, using a 512 bit data frame, the first 32 bits of the 512 bit data frame can be repeated after transmission of the 512 bits, resulting in a total transmission of 544 bits. The receiver then detects the location of the starting bit of that data frame by correlating the locations of the repeating 32 bit data pattern. In another example, bits at the end of the data frame (e.g., 16 bits) are. copied and transmitted before the 512 bit data frame signal, and bits at the beginning of the data frame (e.g., 16 bits) are copied and transmitted after the 512+16=528 bit signal. This technique provides two sets of repeating bit patterns on either end of the 512 bit data frame signal. In this example, the transmission of a 512 bit data frame requires the transmission of 512+16+16=544 bits. Repetition of bits at the beginning of each data frame and/or at the end of each data frame are known collectively as “prefix extension” or “cyclic extension”.
Conventionally, frame shaping has not been performed in an IBOC hybrid FM DAB system. In the absence of any frame shaping, digital frames in each of the separate digital channels A and B in conventional DAB systems must be synchronized so that correlating data frames, e.g., 512 bit data frames, are applied to an Inverse Fast Fourier Transform (IFFT) processing block at the same time. Prefix extension relaxes the accuracy required in OFDM frame synchronization. However, digital samples in the respective digital frames of the two digital channels A and B relating to samples from the same IFFT processing should be passed to a Fast Fourier Transform (FFT) block in the receiver at the same time.
Utilizing frame shaping, the following analysis is used. Let us assume that the output complex samples of an IFFT block are t0, t1, . . . , t511. The transmitted samples, therefore, will be t0, t1, . . . , t511, t512, . . . , t543, where t512=t0, t513=t1, . . . , t543=t31, due to cyclic extension. Note that ti*t*(i+512)=E+j0=E, for i=0, 1, . . . , 31, where E is the energy per transmitted symbol. This is due to the fact that ti=t(i+512) so that there will only be a real result from the complex multiplication equal to the transmitted symbol energy. This is not true for other i's because there is no replica of them transmitted within the cyclic extension. In other words, the samples that have been cyclically extended add up in amplitude while the rest add up in power under a correlation process.
FIG. 7 depicts the integration of a plurality of cyclically extended data frames to provide a long-term tracking function to locate the starting point of subsequent data frame signals.
In particular, in the absence of channel impairments, there would presumably be no Quadrature-phase Q output at point A as shown in FIG. 7, and a relatively larger In-phase I output will result at point A as well. However, in the presence of channel impairments, the output levels at point A due to a similarity in samples from one OFDM symbol to the next can change. This makes a simple threshold approach useless, and thus the starting bit of an OFDM data frame must be determined as an average over several OFDM data frames as depicted in FIG. 7. Accordingly, FIG. 7 shows the use of N stages 1, 2, . . . N, each comprising a delay of, e.g., 544 bits or samples, and a summer, to allow an ‘integration’ of the starting bit of a data frame in the sliding window integrator 702. The delays for averaging are shown in the example as having a length of 544 bits (or samples) based on the length of the cyclically extended data frame. After 544 bits (or samples) of a particular OFDM data frame (or symbol) we are at the same point in the following OFDM data frame (or symbol).
Point B, after the N stages of delay and summation, represents the summation of the contribution of N OFDM data frames (or symbols) to gain averaging power.
Now, since the contribution from the, e.g., 32 bits (or samples) that have been cyclically extended result in a higher real term output, we can locate the beginning of the OFDM data frame (or symbol) by integrating the output at point B shown in FIG. 7 over, e.g., 32 bits (or samples).
The integration is represented by a sliding window integrator 702 in FIG. 7. Upon integration, there should be a peak at the sample time corresponding to the last cyclically extended bit (or sample). In practice, integration is performed using a windowing function to minimize the effect of the outputs due to non-cyclically extended bits (or samples) in the process. A simple windowing function may be implemented to shorten the integration period to a value less than, e.g., 32 bits (or samples).
The output at point C in FIG. 7 is a random looking signal having periodic peaks, as shown in FIGS. 8A and 8B. FIGS. 8A and 8B show an example of both the In-phase (I) and quadrature-phase (Q) portions of an OFDM data frame in the time domain.
In FIGS. 8A and 8B, signal peaks occur relating to a cyclically extended portion 802a, 802b (FIG. 8A) and 804a, 804b (FIG. 8B), e.g., every 544 bits (or samples). However, each peak 802, 804 will take on a random value due to the dynamic nature of the channel. Thus, a tracking loop 704 selects the random-valued peaks 802, 804 in a way which does not lose coherence from OFDM data frame to data frame if, e.g., the peaks disappear due to severe fades.
The output of the tracking loop 704 at point D provides a periodic rectangular pulse with a duty cycle of, e.g., 1/544 at a constant peak value. A pulse train output at point D indicates the starting bit (or sample) of a received OFDM frame signal, e.g., for passage to a FFT engine in a receiver.
Such a conventional OFDM frame synchronization technique works fine in the absence of any first adjacent channels, i.e., channels occupying bandwidth immediately adjacent to that used for the transmission of the OFDM frame signals of interest. Unfortunately, in the presence of a first adjacent channel, one of the outer subcarriers (i.e., one of the digital channels A or B) will suffer from severe interference with the analog channel C of the first adjacent channel. This interference will likely cause detection and correlation of the cyclically extended bits (or samples) in the OFDM channel of interest to fail.
Interference between a first adjacent OFDM channel and the OFDM channel of interest is shown in FIG. 9.
In particular, FIG. 9 shows an OFDM channel of interest comprising digital broadcast signals A and B and an analog broadcast signal C. However, the analog broadcast signal CI of a first adjacent OFDM channel overlaps one of the digital broadcast signals A. In FIG. 9, the first adjacent interfering channel is shown at the edge of coverage of the OFDM channel of interest, which is where the interference is typically the worst.
OFDM frame synchronization becomes seriously affected during the duration of the severe interference caused by the adjacent first channel with respect to the digital broadcast channel A of the OFDM channel of interest, as shown in FIG. 9, because detection of the cyclically extended bit patterns may not be adequately performed, even as averaged over a plurality of frames, because of the ever-presence of an interfering adjacent first channel.
There is thus a need for an improved OFDM frame synchronization technique capable of reliably detecting the location of a starting point of an OFDM data frame in the presence of an interfering first channel.